Accurate prediction of global inclined irradiance (GII) is essential for optimizing solar energy systems across diverse climatic zones. Nevertheless, existing empirical and machine learning models often fail to capture the complex spatiotemporal dynamics governing solar radiation. This study developed and evaluated a hybrid deep learning framework integrating a Convolutional Neural Network (CNN) with a Bidirectional Gated Recurrent Unit (BiGRU) network to predict GII using readily available meteorological inputs. The hybrid architecture was trained and validated using datasets from four stations in the Northern Territory representing tropical, semi-arid, and arid climates. Different performance metrics were used to evaluate the model, and the results were compared with those of classical models and previous studies. Global horizontal irradiance (GHI) emerged as the dominant predictor, while temperature and humidity contributed secondary, climate-dependent effects. The proposed CNN-BiGRU model achieved exceptional predictive accuracy, with R2 values ranging from 0.986 to 0.998 and RMSE values ranging from 23.531 to 66.456 W/m2, outperforming conventional CNN, GRU, and BiLSTM models by up to 41.61% and empirical models by 35%-80%. These results demonstrate that coupling convolutional and bidirectional temporal layers yields physically consistent and highly generalizable GII forecasts, providing a robust foundation for enhancing solar energy modeling and supporting the global transition toward renewable energy systems.